Evidence That Both Growing DNA Chains at a Replication Fork Are Synthesized Discontinuously?

Transcription

1 STERNGLANZ, WANG. AND DONEGAN J. Biol. Chem. 250, Schiffman, S., Theodor,., and Rapaport, s.. (1969), Biochemistry, Shapiro, S. S., and Waugh, D. F. (1966), Thromb. Diath. Haemorrh. 16, 469. Stenn, K. S., and Blout, E. R. (1972), Biochemistry, Swank, R. T., and Munkres, K. D. (1971), Anal. Biochem. 39, 462. Evidence That Both Growing DNA Chains at a Replication Fork Are Synthesized Discontinuously? Rolf Sternglanz,* Helen F. Wang, and James J. Donegan ABSTRACT: Escherichia coli, Bacillus subtilis, and T7-infected E. coli have been labeled with short pulses of [3H]thymidine, and the labeled DNA has been examined by sedimentation in alkaline sucrose. n all three systems, the great majority of the DNA labeled by a short pulse is found in the form of small DNA chains of los, the so-called Okazaki pieces. The B. subtilis and T7 nascent DNA fragments hybridize with equal efficiency to the separated strands of B. subtilis and T7 DNA, respectively. The results suggest that both growing DNA chains at a given replication fork are synthesized discontinuously in the case of E. coli, B. subtilis, and T7. We have found that the method used to terminate the pulse affects the size distribution of the labeled DNA; some methods allow joining of nascent DNA fragments after termination of the pulse. Previous reports of discontinuous DNA synthesis on only one growing DNA chain and continuous synthesis on the other DNA chain are probably due to preferential joining of Okazaki pieces on the DNA chain growing in the overall 5' - 3' direction. The well-known work of Okazaki and co-workers has shown that short pulses of [3H]thymidine are incorporated into small DNA chains, the so-called Okazaki pieces, in a variety of bacterial and bacteriophage systems (Okazaki et al., 196a,b). Pulse-chase experiments have demonstrated that the Okazaki pieces are the precursors of long DNA chains; the latter are indistinguishable in length from the bulk of the DNA. Thus the idea has arisen that one or both of the two growing DNA strands at a replication fork are synthesized discontinuously, and later joined together by DNA ligase. The question of whether one or both growing DNA chains at a given replication fork are synthesized discontinuously is still not settled. Okazaki et ai. (196a,b) reported that label from a very short pulse appears almost exclusively in small pieces in E. coli and T4, implying that both strands are made discontinuously. n three bacteriophage systems, T4, A, and SPP1, Okazaki pieces hybridize equally to both of the separated DNA strands (Sugimoto et al., 1969; Ginsberg and Hurwitz, 1970; Polsinelli et ai., 1969). However, since DNA is now known to replicate bidirectionally in many systems, hybridization data alone cannot be definitive in settling the quest ion. There have been reports in the case of E. coli (yer and Lark, 1970; Louarn and Bird, 1974) and B. subtilis (Okazaki et al., 1970; Kainuma and Okazaki, 1970) that approximately half the label in a short pulse is found in large DNA chains. DNA hybridization studies as well as other methods have led these workers to conclude that only one DNA strand at a replication * From the Department of Biochemistry, State University of New York, Stony Brook, New York Receiued August Supported by grants from the National nstitutes of Health (A 119) and the American Cancer Societ) (NP 142). Part of this work was presented at the CN-UCLA Symposium on DNA Replication. Squaw Valley, California, March fork is synthesized discontinuously in the form of Okazaki pieces, while the other strand is synthesized continuously. n this paper we report on pulse-labeling studies of E. coli, B. subtilis, and T7-infected cells. n all cases, the great majority of the DNA labeled by a short pulse under our conditions is in the form of small DNA fragments, i.e., Okazaki pieces. We have found that the method used to terminate the pulse affects the size distribution of labeled DNA and that some methods allow joining of nascent DNA fragments after termination of the pulse. Previous reports of discontinuous DNA synthesis on only one DNA chain and continuous synthesis on the other DNA chain are probably due to preferential joining of Okazaki pieces on the DNA chain growing in the overall 5' - 3' direction. Materials and Methods Bacterial Strains. E. coli strains included strain B, TAU thy- arg- ura-, and LC 434 F- thy- leu- Bl- X Ximm 434 (Louarn and Bird, 1974). The B. subtilis strain was 16 thytrp- (Wilson et al., 1966). Growth, Pulse Labeling, and Lysis of E. coli. Cells were grown in M9 medium supplemented with 0.4% glucose and 0.1% casamino acids. The medium for E. coli TAU was supplemented with 2 wg/ml thymine, pg/ml uracil, 50 pg/ml arginine, and 0 pg/ml tryptophan. n some cases [4C]thymine was added to the medium to uniformly label the DNA (0.05 pci/ml final concentration). The medium for E. coli LC 434 was supplemented with 2 wg/ml thymine, 25 pg/ml leucine, and 1 Mg/ml thiamine. Cells were grown at 37 "C to 1.5 X los cells/ml, switched to 25 OC, and grown to 2.5 X 1 O cells/ml, diluted 1 : 1 with fresh medium and again grown at 25 O C to 2.5 X O cells/ml. Five-milliliter aliquots of the culture were pulse labeled at 25 O C by the addition of 50 p1 of 1 mci/ml [3H]thymidine (7 Ci/mmol). Pulses were terminated with ml of - 10 OC acetone. The cells were pelleted, 13 BOCHEWSTRY, VOL, NO

2 DSCONTNUOUS DNA REPLCATON washed in 6 ml of cold 0.02 M KCN, 0.02 M EDTA, ph 7.5, and re-pelleted. They were then resuspended in 0.1 ml of 0.02 M EDTA, ph 7.5, and lysed by the addition of 0.5 ml of 0.3 M NaOH, 0.02 M EDTA, and 50 p1 of 10% sarkosyl (Geigy ndustrial Chemicals). The lysates were placed in boiling water for 2 min and cooled quickly in ice-water. Growth, Pulse Labeling, and Lysis of B. subtilis. B. subtilis 16 thy- trp- was grown in a minimal medium described previously (Wang and Sternglanz, 1974). n some cases, [ T thymine was added to the medium to uniformly label the DNA (0.05 pci/ml final concentration). Cells were grown at 37 OC to 1.5 X lo cells/ml, switched to 25 OC, and grown to 2 x Os cells/ml, diluted 1:l with fresh medium and grown at 25 OC to 3 X lo cells/ml. Ten-milliliter portions of the culture were pulse labeled at 25 OC by the addition of 0.1 ml of 1 mci/ml [3H] thymidine (7 Ci/mmol). Pulses were terminated with 40 ml of - 10 OC acetone. The cells were pelleted, washed in ml of cold 0.02 M KCN, 0.02 M EDTA, ph 7.5, and re-pelleted. They were then resuspended in 0.1 ml of 0.02 M EDTA, % sucrose. Ten microliters of 1 mg/ml lysozyme was added and the mixture incubated at 37 OC for 10 min. The cells were lysed by the addition of alkali and sarkosyl exactly as described above for E. coli. Growth, Pulse Labeling, and Lysis of T7-nfected Cells. E. coli B was grown as described above. Five-milliliter portions of the culture at 2.5 X lo cells/ml were infected at 25 OC with T7 at moil = 5. Twenty minutes after infection the cells were pulse labeled for 7 s and lysed exactly as described above for un-infected E. coli. Alkaline Sucrose Gradient Sedimentation. The lysates described above were layered on top of linear 5% (w/v) alkaline sucrose gradients containing 0.3 M NaOH, 0.5 M NaCl, 0.01 M EDTA. At the bottom of each sucrose gradient (10. ml) was a 1-ml CsCl shelf (p = 1.74 g/cm3; 5. g of CsCl dissolved in 4.2 ml of 0.3 M NaOH, 0.5 M NaC1, 0.01 M EDTA). Sedimentation was in a Beckman L2-65B ultracentrifuge for 16 h at rpm using an SW41 rotor and cellulose nitrate tubes. Centrifuge tubes were fractionated from the top and the acid-precipitable radioactivity in each fraction determined as described previously (Wang and Sternglanz, 1974), except that 17 equal fractions were collected from each tube. Also, the bottom of each tube was washed with 1 ml of 0.3 M NaOH, 0.5 M NaCl, 0.01 M EDTA, and then TCAprecipitated. This was called the 1th fraction of each gradient. Some gradients contained 32PP-labeled denatured DNA sedimentation markers. SV40 DNA converted to the linear form by the Eco R1 restriction endonuclease (Mulder and Delius, 1972), a gift of C. Mulder, and T7 DNA, a gift of F. W. Studier, have sedimentation coefficients in alkali of 16s and 37S, respectively. DNA Hybridization. The procedure for DNA hybridization was essentially as described by Denhardt (1966). T7 DNA and r strands were separated according to Summers and Szybalski (196). The B. subtilis DNA separated strands were a gift of R. Rudner (Rudner et a]., 196). Results Pulse Labeling of E. coli. n order to re-examine the question of whether one or both growing DNA strands at a replication fork are synthesized discontinuously in bacteria, we wanted a DNA pulse-labeling procedure designed to minimize nuclease or ligase action after termination of the pulse. Also, it seemed desirable to label cells at the same temperature as Abbreviation used: moi, multiplicity of infection O 5 c a 0 1 a T -? LL 0 z O 5 T A 6 S 37 s J. J 16s 37 s 1 FRACTON NO V % 40 - a V 30 - a - O - s 10 FGURE 1: Alkaline sucrose gradient sedimentation of pulse-labeled DNA from E. coli B and E. coli TAU. Cells were pulse-labeled with [3H]thymidine at 25 OC and the pulse terminated with -10 C acetone as described in Materials and Methods. (A) E. coli B, -s pulse (0-0) total cpm. (B) E. coli TAU, 13-s pulse, 3H (X-X), 2 67 total cpm; 4C (0-- -0) total cpm. had been used for growth of the cells, rather than to shift the temperature just before the pulse. Our procedure is summarized below; details are given in Materials and Methods. Cells are grown exponentially at 37 OC, diluted with fresh medium, and then grown at 25 O C for about two generations before pulse labeling with [3H]thymidine. The pulse is terminated with - 10 OC acetone, and the cells are washed in 0.02 M KCN, 0.02 M EDTA, and finally lysed with alkali and sarkosyl. The lysates are immediately sedimented through alkaline sucrose gradients. Figure 1 presents the results of pulse labeling two E. coli strains, B and TAU, for and 13 s, respectively, using the procedure described above. t can be seen that the great majority of the pulse-labeled DNA sediments at the standard position for Okazaki pieces, about 10s. Specifically, the fraction of the total radioactivity in Okazaki pieces (<S, in fractions 2-) is 66% for E. coli B and 75% for E. coli TAU. For both strains, the fraction of the total radioactivity in large DNA (>40S, at the shelf in fractions 16-1) is only 4%. n the experiment using strain TAU (Figure lb), the DNA was uniformly labeled with [4C]thymine. t can be seen that most of the uniformly labeled DNA is at the shelf and virtually none is at the position of Okazaki pieces. This shows that our lysis BOCHEMSTRY, VOL., NO. 9,

3 ~ ~~ STERNGLANZ. WANG, AND DONECAN Table : Pulse Labeling of Escherichia coli LC 434 under Various Conditions.u % of Total cpm % of Total cpm Growth Pulse Time in Fractions 2- in Fractions 16-1 Expt Conditionsb Stop Solution (s) Total cpm (Okazaki Pieces) (Large DNA) 1 Medium A, 25 "C Medium A, 25 "C 2 Medium A, 25 OC Medium A, 25 OC 3 Medium B', 37- OC Medium B', 37- OC 4 Medium A, 37- "C Medium A, 37- OC Medium A, 37- O C Medium B, 37- OC Medium B, 37- OC Medium B, 37- "C C d e O 04 0 Cells were grown, pulse labeled, and lysed as summarized in the table, and described in Materials and Methods. The lysates were sedimented through alkaline sucrose gradients, and the fraction of the total radioactivity in Okazaki pieces and in large DNA determined. See Figures 1 and 2 for typical sucrose gradients. Medium A is M9 supplemented with 2 pg/ml thymine and 0.1% casamino acids. Medium B' is supplemented with 4 pg/ml thymine and 0.1% casamino acids. Medium B is supplemented with 4 pg/ml thymine and 0.5% casamino acids (Louarn and Bird, 1974). n all cases, 25 kg/ml leucine and 1 pg/ml thiamine were also present. Figure 2A. Figure 2B. e Figure 2C O FRACTON NO. FGURE 2: Alkaline sucrose gradient sedimentation of pulse-labeled DNA from E. coli LC 434. (A) Cells pulse labeled for s at 25 O C and the pulse terminated with -10 "C acetone as described in Materials and Methods, 5343 total cpm. (Band C) Cells grown at 37 OC, pulse-labeled at OC, and the pulse terminated with 10% pyridine, 1 mm KCN as described by Louarn and Bird (1974). (B) 5 pulse, 10 total cpm. (C) 30-s pulse, O 04 total cpm. procedure does not fragment the DNA. The results presented in Figure 1 are in agreement with Okazaki's work on E. coli (Okazaki et al., 196a,b) and clearly suggest that both strands are synthesized discontinuously at a given replication fork. On the other hand, a recent paper by Louarn and Bird (1974) concluded that E. coli DNA is synthesized discontinuously on only one strand, at least for pof A+ strains. We have examined pulse-labeled DNA from the strain used by Louarn and Bird, E. coli K, strain LC 434 (a gift of R. Bird). Figure 2A shows the labeling pattern for LC 434 pulsed for s under our usual conditions. The result is similar to that seen for E. coli B and E. coli TAU; 67% of the radioactivity is in Okazaki pieces (fractions 2-) and only 10% is in large DNA (fractions 16-1). Louarn and Bird reported a different labeling pattern for strain LC 434. They found that a large fraction of the DNA labeled during a 30-s pulse at O C sedimented as intermediate size and large DNA; even with a 5-s pulse at OC at least 40% of the DNA sedimented with S > in their hands. Their procedure differed from ours in three major ways: (1) their supplemented M9 growth medium was slightly richer than ours (generation time was 30 min rather than 35 min at 37 "C); (2) they grew the cells at 37 O C, shifted them to O C for min, and then pulsed at "C; and (3) they terminated pulses with 10% pyridine, 1 mm KCN, and washed the cells in the same solution. When we use the procedure of Louarn and Bird to label strain LC 434 we obtain results similar to theirs. Figures 2B and 2C show - and 30-s pulses for strain LC 434 labeled using their conditions. A significant amount of radioactivity is seen in large DNA (fractions 16-1 ) for both pulses. n order to determine whether the growth medium, the temperature, or the solution used to terminate the pulse (hereafter called the stop solution) was causing the difference in the labeling pattern (contrast Figure 2A with Figures 2B and 2C), we systematically varied the labeling procedure. Table summarizes the results of four experiments. The quantities to be compared are the percent of the total radioactivity found in Okazaki pieces (<S) and in large DNA (>40S) under different conditions. Table clearly shows that the stop solution is an important variable. Although the fraction of labeled DNA found in Okazaki pieces does not depend on the stop solution, the fraction of the pulse-labeled DNA sedimenting to the shelf is always greater in pulses terminated with pyridine than in comparable pulses terminated with acetone. On the other hand, the growth medium does not seem to be a significant variable (experiment 4, Table ; compare medium A with medium B). Comparison of experiments 1 and 2 (cells grown and pulsed at 25 "C) with experiments 3 and 4 (cells grown at 37 "C and pulsed at "C) suggests that shifting the temperature from 37 to OC before pulse labeling also leads to a somewhat greater fraction of large DNA at the shelf. n conclusion, the data of Figure 2 and Table show that strain LC 434 gives results similar to those with other E. coli strains when pulse-labeled under our conditions; most of the radioactivity is found in Okazaki pieces (<S). As argued in detail in the Discussion, we believe that the acetone stop solution and the KCN-EDTA washing of the cells inhibit joining of nascent DNA fragments after termination of the pulse. n contrast, pyridine and KCN do allow some joining to occur. Pulse Labeling of Bacillus subtilis and Hybridization of 140 BOCHEMSTRY, VOL., NO. 9, 1976

4 DSCONTNUOUS DNA REPLCATON 16s 37 s 6 S 37 s J..1 LL 0 O B 6 S 37 s.1.1 T i FRACTON NO. FGURE 4: Alkaline sucrose gradient sedimentatian of pulse-labeled DNA from T7-infected E. coli B. Cells were grown at 25 OC to 2.5 X Os cells/ ml, infected with T7 at moi = 5, and incubation continued at 25 OC. Twenty minutes after infection the cells were pulse labeled for 7 s, the pulse was terminated with -10 "C acetone, and the cells were lysed as described in Materials and Methods. Six individual sucrose gradients, each containing the lysate from 5 ml of T7-infected cells, were fractionated into one set of tubes. An aliquot of each tube was precipitated with % trichloroacetic acid to determine the radioactivity. The rest of the labeled DNA was used for the hybridization experiment shown in Table 111 (experiment ). FRACTON NO. FGURE 3: Alkaline sucrose gradient sedimentation of pulse-labeled DNA from B. subtilis 16 thy- trp-. Cells were pulse labeled at 25 O C and the pulse terminated with -10 O C acetone as described in Materials and Methods. (A) -s pulse, 3H (X-X), 1655 total cpm; 4C (0-- -O), 74 total cpm. (B) -s pulse (e-.), 690 total cpm. the Nascent DNA with Bacillus subtilis DNA Separated Strands. Okazaki and co-workers have reported that, when B. subtilis is labeled with short pulses of [3H]thymidine, some of the radioactivity is found in 10s fragments and the rest is found in large DNA (Okazaki et al., 1970; Kainuma and Okazaki, 1970). When they hybridized the labeled DNA with the separated strands of B. subtilis DNA, they found that the 10s pieces hybridized preferentially to the L strand and the large DNA hybridized preferentially to the H strand. Okazaki and co-workers, therefore, concluded that B. subtilis DNA is replicated discontinuously off the L template strand and continuously off the H template strand. However, in our hands the same results are obtained with B. subtilis as with E. coli. Figure 3 shows the results of - and -s pulses of B. subtilis labeled using the conditions described in this paper (25 'C growth and labeling, acetone stop solution, KCN-EDTA washing of the cells). t can be seen that the great majority of the radioactivity incorporated during a short pulse labels Okazaki pieces and very little radioactivity is found in large DNA. On the other hand, the majority of the uniformly labeled DNA sediments to the shelf (Figure 3A). The results suggest that both growing DNA chains at a replication fork can be replicated discontinuously in B. subtilis. f both DNA chains are replicated discontinuously, then the Table 11: Hybridization of Bacillus subtilis Nascent DNA with the Separated Strands of Bacillus subtilis DNA.a % Hybridization to ExDt L Strand H Strand a Both experiments involved a -s pulse at 25 OC. The distribution of radioactivity was very similar to that seen for the -s pulse in Figure 3. Fractions from the peak of radioactivity in the sucrose gradient were pooled and used for hybridization (Denhardt, 1966). Experiment 1 involved the use of 0 ml of cells concentrated to 5 ml before pulse labeling as described previously (Wang and Sternglanz, 1974). Experiment 2 used unconcentrated cells. Percent hybridization = [(cpm remaining on filter)/(cpm added to filter)] X 100%. Okazaki pieces should hybridize to both of the separated strands of B. subtilis DNA. Table 1 shows that indeed the Okazaki pieces do hybridize with equal efficiency to both the L and H strands of B. subtilis DNA. Pulse Labeling of T7-nfected E. coli and Hybridization of the Nascent DNA with T7 DNA Separated Strands. When T7-infected cells are pulse labeled for 7 s at min after infection, the labeling pattern depicted in Figure 4 is observed. The main peak of radioactivity again is at the position of Okazaki pieces, at about OS, although the peak is broader than is seen in the case of uninfected E. coli or B. subtilis. Very little radioactivity is seen at fraction 14, the position of unit length T7 DNA single strands. We have used the labeled DNA from the 7-s pulse shown BOCHEMSTRY, VOL., NO. 9,

5 STERNGLANZ, WANG, AND DONEGAN Table ll: Hybridization of T7 Nascent DNA with the Separated Strands of T7 DNA.O % Hybridization to Expt Pooled Fractions strand r strand -- b a Both experiments involved a 7-s pulse of T7-infected cells min after infection. Cells were lysed and sedimented in alkaline sucrose gradients. Figure 4 shows the distribution of radioactivity for experiment 1. A similar distribution was seen in experiment 2. n both cases, the indicated fractions from the sucrose gradients were pooled and used for hybridization. Percent hybridization = [(cpm remaining on filter)/(cpm added to filter)] X 100%. Figure 4. in Figure 4 for hybridization to the separated strands of T7 DNA. Fractions 3-5 from the slower sedimenting half of the radioactive DNA peak were pooled for hybridization studies, as were fractions 6- from the rapidly sedimenting half of the peak. Experiment 1 in Table 11 shows that there is approximately equal efficiency of hybridization to the l and r strands of T7 DNA for both pooled fractions 3-5 and pooled fractions 6-. Table 11 also shows another hybridization experiment involving DNA labeled in a 7-s pulse. Radioactive DNA from a sucrose gradient similar to the one shown in Figure 4 was divided into three parts: (1) the slower sedimenting DNA (fractions 2-4); (2) the peak fraction (No. 5); and (3) the more rapidly sedimenting DNA fragments (fractions 6-). Again the labeled DNA hybridizes equally with the 1 and r strands. n summary, the pulse-labeling pattern seen in Figure 4 as well as the hybridization results in Table 111 suggest that T7 DNA is replicated discontinuously on both growing DNA chains of a replication fork. Discussion We have re-examined the question of whether one or both growing DNA chains at a replication fork are synthesized discontinuously. Under our conditions, the majority of the radioactivity incorporated into DNA during a short pulse is found in Okazaki pieces. The fraction of pulse-labeled DNA in short pieces (S < ) is about 70% for E. coli (Figures 1 and 2A) and more than 90% for B. subtilis (Figure 3). On the other hand, most of the uniformly labeled E. coli or B. subtilis DNA sediments to the shelf (fractions 16-1 of the sucrose gradients) under these conditions. f one of the two growing DNA strands at a replication fork is replicated continuously, then one-half of the radioactivity incorporated into DNA during a short pulse should be in large DNA. Presumably the DNA will be as large as the bulk of the DNA. Our evidence is clearly inconsistent with such a model. Even workers who have suggested that one strand is replicated continuously never find as much as 50% of the pulse label in large DNA (yer and Lark, 1970; Louarn and Bird, 1974; Okazaki et al., 1970; Kainuma and Okazaki, 1970). nstead they find a substantial amount of labeled DNA sedimenting at intermediate positions, larger than Okazaki pieces but smaller than bulk DNA. Since the fraction of radioactivity in intermediate-size pieces plus that in large DNA (i.e,, all the DNA larger than s) often approach 50%, both of these size classes have previously been attributed to the strand made continuously. n the case of E. coli, we find approximately -30% of the labeled DNA in the intermediate size class (greater than s but less than 40s; fractions 9- of Figures 1 and 2A). This DNA could either be due to: (1) joining of 10s Okazaki pieces either before or after termination of the pulse, or (2) a rapid rate of DNA polymerization relative to initiation of Okazaki pieces on the strand growing in the overall 5-3 direction. The latter case would lead to some chains longer than a single Okazaki piece, a situation known to occur during in vitro DNA replication in the cellophane disc system (Olivera and Bonhoeffer, 1972). However, in our view this still should be considered discontinuous synthesis, Our results show that the method of stopping the pulse and washing the cells affects the fraction of labeled DNA which is larger than Okazaki pieces (S > ). Previous workers who observed about half the pulse-labeled DNA in pieces larger than s (and who therefore concluded that one strand is synthesized continuously) all used pyridine and/or KCN to stop the pulse and wash the cells (yer and Lark, 1970; Louarn and Bird, 1974; Okazaki et al., 1970; Kainuma and Okazaki, 1970). Comparison of the two methods shows that, while the fraction of labeled DNA found in Okazaki pieces is not significantly different in the two cases, the fraction of labeled DNA found at the shelf is always greater in the case of pyridine-kcn (Figure 2 and Table ). Our results suggest that the pyridine-kcn stop solution allows some joining of intermediate-size pieces (which we consider to be products of discontinuous synthesis as argued above), and that acetone and/or KCN-EDTA used in our procedure prevent much of this joining. Louarn and Bird (1 974) concluded that one strand is synthesized continuously in E. coli, and they ingeniously demonstrated that it is the strand growing in the overall 5-3 direction. They pulse-labeled an E. coli strain lysogenic for phage X and hybridized the labeled DNA to the separated strands of X DNA. Using a pol A+ strain, they found that a short pulse labeled both short and large DNA fragments; the former hybridized preferentially to the l strand and the latter to the r strand of X DNA. Knowing the orientation and position of the integrated X DNA in the E. coli chromosome, they concluded that E. coli DNA is replicated discontinuously on the strand growing in the overall 3-5 direction and is replicated continuously on the strand growing in the overall 5-3 direction. However, in the case of a pol A- strain, they found mainly small DNA after a short pulse, and concluded that both strands are replicated discontinuously. Our results (Figure 2 and Table ) using the same pol A+ strain as Louarn and Bird used lead us to reinterpret their results. We conclude that both strands are replicated discontinuously, and that there is preferential joining of intermediate size nascent DNA fragments on the strand growing in the overall direction under their experimental conditions. Preferential joining of nascent DNA fragments of the strand growing in the overall direction might be expected since these pieces have a greater probability of being immediately adjacent to each other than do pieces on the side of the fork; the latter are initially separated by a large gap. Additional support for this point of view comes from the results of Louarn and Bird with the pol A- strain. Since the only known defect of pol A mutants in DNA replication is in the joining of Okazaki pieces (Kuempel and Veomett. 1970). 142 BOCHEMSTRY, VOL., NO. 9, 1976

6 DSCONTlNUOUS DNA REPLCATON it seems reasonable to conclude that the difference between pol A+ and pol A- strains found by Louarn and Bird (1974) is simply due to the difference in the ability of the two strains to join Okazaki pieces. Okazaki and co-workers have recently come to a similar conclusion in the case of phage P2 (Kainuma-Kuroda and Okazaki, 1975; Kurosawa and Okazaki, 1975). n the case of T7, we also find that the DNA labeled during a short pulse sediments as 10s fragments, but the size distribution is broader than in the case of un-infected E. coli (Figure 4). T7 DNA is known to replicate in the form of unit length linear molecules. Replication is bidirectional, from an origin 17% from the left end of the molecule (Dressler et al., 1972). Concatemers are formed later in infection (Schlegel and Thomas, 1972) but there is no direct evidence that concatemers themselves are replicative forms. f only unit length molecules replicate and if the origin is 17% from the left end, then if one strand is replicated continuously we would expect an asymmetry in the hybridization of the labeled DKA to the separated 1 and r strands. Knowing the polarity of the strands relative to the genetic left end of the molecule (Summers and Szybalski, 196), we would expect the smaller labeled DNA to hybridize preferentially to the 1 strand and the larger labeled DNA to hybridize preferentially to the r strand. Table 11 shows that all size classes hybridize with equal efficiency to both the and r strands. Thus, the data of Figure 4 and Table 11 are consistent with discontinuous replication of both strands of T7 DNA. We therefore conclude that both growing DNA strands at a replication fork are replicated discontinuously in the case of E. coli, B. subtilis, and T7. Our conclusion is based on two assumptions: (1) [3H]thymidine labels both growing DNA chains with approximately equal efficiency; and (2) there is no unusual fragility of the newly replicated DNA. Acknowledgments We thank Aviva Mukamel and Judith P. Baird for technical assistance, Care1 Mulder and F. William Studier for DNA sedimentation markers, and Rivka Rudner for the separated strands of B. subtilis DNA. References Denhardt, D. (1966), Biochem. Biophys. Res. Commun. 23, 641. Dressler, D., Wolfson, J., and Magazin, M. (1972), Proc. Natl. Acad. Sci. U.S.A. 69, 99. Ginsberg, B., and Hurwitz, J. (1970), J. Mol. Bioi. 52, 265. yer, V., and Lark, K. (1970), Proc. Natl. Acad. Sci. U.S.A. 67, 629. Kainuma, R., and Okazaki, R. (1970), J. Jpn. Biochem. SOC. 42, 464. Kainuma-Kuroda, R., and Okazaki, R. (1979, J. Mol. Biol. 94, 213. Kuempel, P., and Veomett, G. (1970), Biochem. Biophys. Res. Commun. 41, 973. Kurosawa, Y., and Okazaki, R. (1975), J. Mol. Biol. 94, 229. Louarn, J., and Bird, R. (1974), Proc. Natl. Acad. Sci. U.S.A. 71, 329. Mulder, C., and Delius, H. (1972), Proc. Natl. Acad. Sci. U.S.A. 69, 32. Okazaki, R., Okazaki, T., Sakabe, K., Sugimoto, K., Kainuma, R., Sugino, A,, and watsuki, N. (196a), Cold Spring Harbor Symp. Quant. Biol. 33, 9. Okazaki, R., Okazaki, T., Sakabe, K., and Sugino, A. (196b), Proc. Natl. Acad. Sci. U.S.A. 59, 59. Okazaki, R., Sugimoto, K., Okazaki, T., mae, Y., and Sugino, A. (1970), Nature (London) 22, 223. Olivera, B., and Bonhoeffer, F. (1972), Nature (London), New Biol. 240, 233. Polsinelli, M., Milanesi, G., and Ganesan, A. (1969), Science 166, 243. Rudner, R., Karkas, J., and Chargaff, E. (196), Proc. Natl. Acad. Sci. U.S.A. 60, 630. Schlegel, R., and Thomas, C. A., Jr. (1972), J. Mol. Biol. 6, 319. Sugimoto, K., Okazaki, T., mae, Y., and Okazaki, R. (1969), Proc. Natl. Acad. Sci. U.S.A. 63, Summers, W., and Szybalski, W. (196), Virology 34, 9. Wang, H., and Sternglanz, R. (1974), Nature (London) 24, 147. Wilson, M., Farmer, J., and Rothman, F. (1966) J. Bact. 92, 16. BOCHEMSTRY, VOL., NO. 9,